Improving Boids Algorithm in GPU using Estimated Self Occlusion

Publication Type:

Conference Paper


Proceedings of SBGames'08 - VII Brazilian Symposium on Computer Games and Digital Entertainment, Sociedade Brasileira de Computação, SBC (2008)





Behavioral models are used in games and computer graphics for realistic simulation of massive crowds. In this paper, we present a GPU based implementation of Reynolds [1987] algorithm for simulating flocks of birds and propose an extension to consider environment self occlusion. We performed several experiments and the results showed that the proposed approach runs up to three times faster than the original algorithm when simulating high density crowds, without compromising significantly the original crowd behavior.